assertion failures under sunos on the sparc

assertion failures under sunos on the sparc

[David S. Miller]

It seems that if you interrupt a build of a tree halfway through, and
make doesn't clean up the half-written .o file properly, later on when
you link the screen fills with a monsterous amount of 'bfd-assertion
failure bfd/sunos.c'.  I get about 400 of them for the linux kernel
when this happens.

The error message is warranted because the linker is processing bogus
information, however this amount of verbosity is very overwhelming to
the user.  Is there a way to inform the user which object file is the
corrupt one instead during examination of objects in bfd?  If so, this
would be more useful than what happen currently.

This is with binutils-2.6

Later,
David S. Miller
davem@caip.rutgers.edu

Re: assertion failures under sunos on the sparc

[Richard Stallman]

    It seems that if you interrupt a build of a tree halfway through, and
    make doesn't clean up the half-written .o file properly,

How come Make doesn't do that?  Make is supposed to delete the current
target if you kill it after that target has been altered.

Re: assertion failures under sunos on the sparc

[David S. Miller]

   Date: Mon, 27 Nov 1995 21:30:28 -0500
   From: Richard Stallman <rms@gnu.ai.mit.edu>
   CC: gas2@cygnus.com

       It seems that if you interrupt a build of a tree halfway through, and
       make doesn't clean up the half-written .o file properly,

   How come Make doesn't do that?  Make is supposed to delete the current
   target if you kill it after that target has been altered.

I severly apologize, it now seems that I am getting bfd assertion
failures like mad even for what appears to be correct builds of my
sources.  This is with gcc-2.7.1 and binutils-2.6 on the sparc under
SunOS.

For all I know, make was doing the correct thing.

I will hunt this down some more, it looks pretty bad.

Later,
David S. Miller
davem@caip.rutgers.edu